Superhelical torsion controls DNA interstrand cross- linking by antitumor
cis
- diamminedichloroplatinum(II)
Superhelical torsion controls DNA interstrand cross- linking by antitumor cis - diamminedichloroplatinum(II)
Oldrich
Vrána
,
Vladimír
Boudny
and
Viktor
Brabec*
Institute of Biophysics, Academy of Sciences of the Czech Republic, Královopolská 135, CZ-61265
Brno
,
Czech Republic
Received July 19, 1996;
Revised and Accepted August 28, 1996
ABSTRACT
Negatively supercoiled, relaxed and linearized forms of pSP73 DNA were modified
in cell-free medium by
cis
-diamminedichloroplatinum(II) (cisplatin). The frequency of interstrand cross-links (ICLs) formed in these DNAs has been determined by: (i)
immunochemical analysis; (ii) an assay employing NaCN as a probe of DNA ICLs of
cisplatin; (iii) gel electrophoresis under denaturing conditions. At low levels
of the modification of DNA (<1 Pt atom fixed per 500 bp) the number of ICLs formed by cisplatin was
radically enhanced in supercoiled in comparison with linearized or relaxed DNA.
At these low levels of modification, the frequency of ICLs in supercoiled DNA
was enhanced with increasing level of negative supercoiling or with decreasing
level of modification. In addition, the replication mapping of DNA ICLs of
cisplatin was consistent with these lesions being preferentially formed in negatively supercoiled DNA between guanine residues in both the 5
'
-d(GC)-3
'
and the 5
'
-d(CG)-3'
sites. Among the DNA adducts of cisplatin the ICL has the markedly greatest
capability to unwind the double helix. We suggest that the formation of ICLs of
cisplatin is thermodynamically more favored in negatively supercoiled DNA owing
mainly to the relaxation of supercoils.
INTRODUCTION
A probable mechanism of anticancer activity of
cis-
diamminedichloroplatinum(II) (cisplatin) involves formation of platinum-DNA adducts which are capable of blocking DNA and RNA synthesis (for
general reviews see
1
-
7
) and induce programed cell death (
8
,
9
). The DNA adducts of cisplatin have been extensively studied, in particular in
linear DNA modified by this drug in cell-free medium to relatively high levels (
r
b
>= 0.001;
r
b
is defined as the number of platinum atoms fixed per nucleotide residue) (
10
,
11
). These studies have revealed that cisplatin forms ~90% 1,2 intrastrand d(GpG) or d(ApG) cross-links in linear DNA; intrastrand cross-links between non-adjacent purine residues, interstrand cross-links (ICLs) and monofunctional adducts were minor
adducts. Considerable evidence suggests that the antitumor efficacy of
cisplatin is a result of the formation of DNA 1,2-intrastrand cross-links by this drug (see for instance
3
-
5
). Nevertheless, DNA ICLs of cisplatin have also been considered lesions
relevant to the biological effects of this drug (see for instance
12
-
16
), but their relative efficacy remains unknown. For instance, recent results (
13
,
17
) suggest that there is a selective increase in gene-specific ICL repair in cisplatin-resistant human ovarian cells. Thus, formation of DNA ICLs of
cisplatin is challenging to understand, not only from a mechanistic view but
also from the therapeutic one. Therefore, it is not surprising that the nature
of the DNA ICLs of platinum complexes and the biological significance of these
lesions are being intensively investigated at present (see for example
18
-
33
).
The formation of cisplatin adducts in DNA can be affected by its conformation (
34
). One of the important factors that influences the conformation of DNA is its
global topology (
35
,
36
). DNA
in vivo
is generally negatively supercoiled in both eukaryotic and prokaryotic cells
(for general reviews see
35
-
37
). Whereas a significant portion of bacterial DNA supercoils is thought to be
unrestrained inside the cell, the superhelicity of eukaryotic DNA is attributed
to wrapping into nucleosomes. In addition, RNA polymerase transiently changes the local level of DNA supercoiling as it transcribes, also generating positive supercoiling. There are well-documented examples of the stabilization of alternative DNA conformations
by supercoiling (
35
-
38
). Interestingly, it was recently shown (
39
) that cisplatin toxicity is influenced by DNA supercoiling in prokaryotic
cells.
The formation of cisplatin adducts in DNA alters its conformation (
3
-
5
,
7
,
23
,
40
-
44
). The conformational changes induced by bifunctional adducts of cisplatin
result in particular in bending and unwinding of the DNA duplex (
4
,
5
,
23
,
26
,
40
,
42
). In general, negative supercoiling tends to stabilize conformations in which
there is unwinding of the DNA (
35
-
37
). Thus, it seems reasonable to assume that some bifunctional adducts of
cisplatin could be more favored in negatively supercoiled DNA (in comparison with relaxed or linear DNAs) than others. In other words, it may be expected
that the individual types of cisplatin adducts could appear in negatively
supercoiled DNA with a different frequency and/or sequence specificity than in
relaxed or linear DNAs or even new types of platinum adducts not allowed in
linear DNA could appear in supercoiled DNA. These expectations are corroborated by recent observations demonstrating that monofunctional cisplatin adducts in negatively supercoiled DNA are
converted to bifunctional lesions more readily than in relaxed DNA (
45
). Also, interestingly, a novel adduct of cisplatin in the sequence 5'-TACT-3', capable of terminating DNA synthesis, has been found in DNA from cisplatin-treated eukaryotic cells, but not in linear DNA modified by this
drug in cell-free medium (
46
).
Here we present experiments which were carried out in cell-free medium to investigate the effect of negative supercoiling on DNA
interstrand cross-linking by cisplatin. We find that superhelical torsion can affect
interstrand cross-linking efficiency of this anticancer drug in plasmid DNA.
MATERIALS AND METHODS
Starting materials
Cisplatin was synthesized and characterized by Lachema (Brno, Czech Republic).
cis-
[Pt(NH
3
)
2
(H
2
O)Cl]
+
was generated from cisplatin as described previously (
18
). Ultrapure acrylamide, bis(acrylamide) and urea were from Merck. The
CircumVent Thermal Cycle Dideoxy DNA Sequencing Kit with Vent
R
DNA polymerase, the Klenow fragment of DNA polymerase I, restriction
endonucleases and calf thymus DNA topoisomerase I were purchased from BioLabs. T4 polynucleotide kinase was from Boehringer. If not stated otherwise, the enzymes were used employing the
buffers and other experimental conditions recommended by the manufacturer. The
primer 5'-d(TAATACGACTCACTATAGGG)-3' was from BioVendor (Brno, Czech Republic). The radioactive products were from Amersham. Plasmid
pSP73 (2464 bp) was prepared as described earlier (
47
). The sample of the native plasmid prepared in this laboratory contained >95%
negatively supercoiled and <5% relaxed forms. Under the present experimental conditions, superhelical
density [sigma] of this naturally supercoiled DNA was -0.063, calculated on the basis of the data on DNA unwinding by
cisplatin (
48
). Plasmid pSP73 DNA was [gamma]-irradiated at a concentration of 40 [mu]g/ml in 5 mM Tris-HCl, 0.5 mM EDTA, pH 7.8, using a cobalt source; other
conditions were as described earlier (
49
,
50
). The nicked DNA samples were subjected to electrophoresis on 1.5% agarose gels
run at 25oC in the dark with TBE buffer with the voltage set at 30 V. The gels were then stained with ethidium bromide and were examined on
Polaroid 667 film using a transilluminator.
Preparation of DNA molecules of varying superhelical density
DNA topoisomers were prepared by relaxation of plasmid pSP73 in the presence of
ethidium bromide or netropsin (
45
,
51
). Twelve micrograms of plasmid were incubated in the dark with 24 U DNA
topoisomerase I in 72 mM KCl, 35 mM Tris-HCl, pH 8.0, 5 mM MgCl
2
, 5 mM dithiothreitol, 2 mM spermidine, 5% glycerol in a final volume of 300 [mu]l also containing different amounts of ethidium bromide (0-30 mM) or netropsin (15-60 mM). The mixtures were incubated for 90 min at 37oC. The reactions were terminated by adjusting the concentration of EDTA to
10 mM. The mean linking number ([Delta]
Lk
) for each negatively supercoiled DNA preparation was determined with respect to
relaxed DNA using a series of chloroquine gels (
52
). The superhelical density, [sigma], was calculated (
53
) as 10.4 * [Delta]
Lk
/
N
, where
N
is the number of base pairs in the plasmid (2464 for pSP73).
Platination reactions
Supercoiled, relaxed or linearized forms of pSP73 DNA were modified by cisplatin
in 10 mM NaClO
4
at 37oC in the dark for 48 h if not stated otherwise. The
r
b
values in these samples were determined by flameless atomic absorption
spectrophotometry (FAAS) using a Unicam 939 AA spectrometer with graphite
furnace.
Immunochemical analysis
Antibodies designated AB
inter
were elicited against poly(dG-dC)[middot]poly(dG-dC) modified by cisplatin at
r
b
= 0.08 in 10 mM NaClO
4
for 48 h at 37oC (
28
,
54
). It was shown in the same way as in the recent paper of Malinge
et al.
(
28
) that these antibodies did not recognize platinum residues, DNA intrastrand or
monofunctional adducts of cisplatin and that recognition is specific for DNA
ICLs of cisplatin. The procedures for immunoenzymatic analysis of the
antibodies and enzyme-linked immunosorbent assay (ELISA) were also as described recently (
55
,
56
).
Sodium cyanide assay
These experiments were performed in the same way as described in a recent paper
(
47
). Briefly, 6.4 [mu]g plasmid pSP73 containing cisplatin adducts and linearized by
Eco
RI before or after the platination reaction (
vide infra
) were treated with 0.2 M NaCN, pH 8.3, for 4 h at 37oC in the dark (the stock solution of 1 M NaCN was prepared in 0.1 M Tris-HCl and the pH was adjusted to 8.3 with 12 M HCl). The final volume of this reaction was 0.5 ml. After 4 h the samples were precipitated with ethanol, the pellets
dissolved in 0.1 M HNO
3
and analyzed for platinum content by FAAS.
Interstrand cross-link assay
If not stated otherwise, cisplatin at varying concentrations was incubated with
2 [mu]g supercoiled pSP73 DNA or with this plasmid linearized by
Eco
RI. The samples were then precipitated with ethanol and the circular DNA already
modified by cisplatin was subsequently linearized by
Eco
RI. The linear duplexes were then analyzed for DNA ICLs in the same way as
described in several recent papers (
18
,
25
,
47
,
57
,
58
). The linear duplexes were first 3'-end-labeled by means of the Klenow fragment of DNA polymerase I
and [[alpha]-
32
P]dATP. The samples were deproteinized with phenol, precipitated with ethanol
and the pellet dissolved in 18 [mu]l 30 mM NaOH, 1 mM EDTA, 6.6% sucrose, 0.04% bromophenol blue. The number of
ICLs was analyzed by electrophoresis under denaturing conditions (
18
,
47
) on alkaline agarose gels (1%). After electrophoresis was completed the bands corresponding to single-stranded DNA and interstrand cross-linked duplex were cut out and their radioactivity quantified on a LKB
Wallac 1410 Beta spectrometer (Finland).
Sequence specificity of cisplatin-DNA adducts
Replication mapping of lesions induced by cisplatin in plasmid pSP73 DNA was
conducted in the way described in detail in a recent paper (
58
). Control or platinated double-stranded DNAs were used as the templates. The platinated templates were prepared by modification by cisplatin at
r
b
= 0.005 of either supercoiled pSP73 DNA ([sigma] = -0.05) or pSP73 DNA linearized (before the platination reaction) by
Nde
I. Supercoiled control (non-platinated) or platinated supercoiled DNAs were cleaved by the two endonucleases
Nde
I and
Hpa
I, while DNA platinated in linearized form was, after the platination reaction, also cleaved by
Hpa
I. The resulting (
Nde
I/
Hpa
I) fragments were deproteinized and the platinated fragments treated with 0.2 M
NaCN in 20 mM Tris-HCl, pH 8.3, for 4 h at 37oC, if not stated otherwise. The 221 bp fragments with and without
ICLs were purified using 2% denaturing agarose gels and used in the mapping experiments (
18
). The 221 bp fragments used in the mapping experiments were also subjected to
restriction analysis using
Bgl
II,
Cla
I or
Hin
dIII. The cleavage products of the (
Nde
I/
Hpa
I) fragment were analyzed by electrophoresis on non-denaturing 15% polyacrylamide [mono:bis (acrylamide) ratio 29:1] gels running at 25oC in the dark with TBE buffer with the voltage set at 45 V. The gels
were then stained with ethidium bromide and were examined on Polaroid 667 film
with a transilluminator.
RESULTS
The initial experiments of the present work were carried out to compare the
numbers of ICLs formed by cisplatin in supercoiled and linear DNAs. In these
experiments, we used plasmid pSP73 (2464 bp, native supercoil density [sigma] = -0.063) which was modified by cisplatin in two ways. One series of
the samples was prepared by modifying the plasmid only after it had been linearized by
Eco
RI (
Eco
RI cuts only once within plasmid pSP73). The samples of the other series were prepared by modification of the supercoiled
plasmid by cisplatin and only after the platination reaction was complete was
the plasmid linearized by
Eco
RI. Thus, we prepared two types of linear DNA molecules of the same length and
nucleotide sequence modified by cisplatin which could be differently affected
by this drug if different DNA topology during the platination reaction plays a
role. The two samples were analyzed for ICLs by different techniques in order
to reveal these differences. It was also verified by means of FAAS that the
amount of cisplatin coordinated to the base residues in DNA was independent of
the DNA topology during the platination reaction in a broad range of
r
i
values (0.0001-0.08) (
r
i
is defined as the molar ratio of free platinum complex to nucleotide phosphates at
the onset of incubation with DNA) and at any reaction time. Importantly, the
modification was complete (i.e. all platinum was bound) at these
r
i
values within 48 h.
Immunochemical analysis
We prepared and characterized antibodies which bind selectively to ICLs formed
by cisplatin in DNA (AB
inter
) in the same way as in the work of Malinge
et al.
(
28
). These antibodies do not recognize platinum residues, intrastrand cross-links or monofunctional DNA adducts of cisplatin. We used AB
inter
to show whether linearized pSP73 DNA, which was modified by cisplatin either in
the linear or native supercoiled form, competitively inhibits binding of these
antibodies to their immunogen in an ELISA. The competitive inhibition was
evaluated in terms of the inhibitor concentration at 50% inhibition of the
antibodies (IC
50
). As shown in Figure
1
, AB
inter
recognized the DNA modified by cisplatin in the supercoiled form better than in
the linearized form. The difference in recognition was increased with
decreasing level of modification. The results of this analysis (Fig.
1
) are consistent with the idea and support the hypothesis that at relatively low
levels of modification by cisplatin (
r
b
< 0.001) ICLs are formed in supercoiled DNA with a higher frequency (the
percentage fraction of platination events resulting in ICLs) than in linear
DNA.
.
Stability of the adducts formed by cisplatin in supercoiled or linearized pSP73
plasmid in the presence of NaCN
a
r
b
* 10
4 b
Pt content
c
(pmol)
Fraction of stable Pt adducts
d
(%)
Total
e
SC
NaCN
f
LIN
NaCN
g
SC
NaCN
f
LIN
NaCN
g
1
2.0
0.8 +- 0.1
0.2 +- 0.1
40
10
2
4.0
1.1 +- 0.2
0.3 +- 0.1
28
8
5
10.0
1.3 +- 0.3
0.7 +- 0.1
13
7
10
20.0
2.4 +- 0.3
1.2 +- 0.2
12
6
a
pH 8.3, 37oC; for other details see text.
b
r
b
before NaCN treatment.
c
The amount of DNA modified by cisplatin was 6.4 [mu]g.
d
The fraction of cisplatin adducts resistant to the NaCN treatment.
e
Platinum content before NaCN treatment.
f
DNA was modified in supercoiled form, linearized and treated with NaCN. The
values represent mean +- SEM of four determinations from two independent experiments.
g
DNA was modified in linearized form and treated with NaCN. The values represent
mean +- SEM of four determinations from two
independent experiments.
NaCN treatment
Most of the cisplatin molecules bound to DNA can be removed by the action of cyanide ions. It has been shown (
47
) that at pH 8.3 and 37oC treatment with NaCN at a concentration of 0.2 M for 4 h can effectively
remove ~90% of 1,2-intrastrand cross-links of cisplatin from double-helical DNA, whereas ICLs are more stable under these
conditions (only ~15% of all ICLs are removed). The samples of linearized plasmid containing
adducts of cisplatin formed when the DNA was either in supercoiled or in
linearized form were treated with 0.2 M NaCN at pH 8.3 and 37oC for 4 h in the dark. After this treatment, the content of platinum was
determined in these samples by FAAS (Table
1
). More than 90% of the platinum was removed from DNA modified in the linear
form. In contrast, significantly less platinum was removed from DNA modified in
the supercoiled form (and subsequently linearized) and the fraction of platinum
not extracted with CN
-
increased with decreasing level of the initial modification (
r
b
). These results are consistent with the immunochemical investigations (Fig.
1
), i.e. they also indicate that: (i) more ICLs were formed in negatively
supercoiled than in linear DNA; (ii) at relatively low levels of platination (
r
b
< 0.001) the frequency of ICLs formed by cisplatin in supercoiled DNA was
enhanced with decreasing level of DNA modification.
DNA interstrand cross-linking
A method for precise and quantitative determination of interstrand cross-linking by cisplatin in DNA has been previously described (
18
,
25
,
47
,
57
,
58
). Upon electrophoresis under denaturing conditions, 3'-end-labeled strands of linearized plasmid pSP73 containing no ICLs migrate as a 2464 base single strand, whereas the
interstrand cross-linked strands migrate more slowly as a higher molecular mass species. The
bands corresponding to more slowly migrating interstrand cross-linked fragments were noticed if cisplatin was used to modify DNA in both
the linearized and supercoiled forms at
r
b
as low as 1 * 10
-4
(Fig.
2
A and B). The intensity of the more slowly migrating band increased with growing
level of modification. The radioactivity associated with the individual bands
in each lane was measured to obtain estimates of the fraction of non-cross-linked or cross-linked DNA under each condition. The frequency of ICLs was
calculated using the Poisson distribution from the fraction of non-cross-linked DNA in combination with the
r
b
values and the fragment size (
57
). As summarized in Figure
2
C, at low levels of modification (
r
b
= 0.0001-0.001) cisplatin showed a noticeably higher interstrand cross-linking efficiency in supercoiled DNA than in its linearized form.
In addition, while the frequency of ICLs formed in the linearized DNA was
increased only very slightly with decreasing
r
b
, the frequency of ICLs formed in supercoiled DNA increased much more radically.
Sequence preference of DNA interstrand cross-linking
It has been shown (
47
) that in linear DNA cisplatin preferentially forms ICLs between guanine
residues in 5'-(GC)-3' base pairs. The results of the present paper (Figs
1
-
3
and Table
1
) indicate that ICLs of cisplatin can be formed in negatively supercoiled DNA
with a higher frequency than in linear or relaxed DNAs. It could be suggested, as one possible explanation of this phenomenon, that cisplatin forms ICLs in negatively supercoiled DNA at more sites and/or
different sequences than in linear DNA. In order to test this hypothesis, we
prepared a 221 bp (
Nde
I/
Hpa
I) restriction fragment which was isolated from DNA modified by cisplatin either
in the supercoiled ([sigma] = -0.05) or the linear form. These fragments modified by cisplatin
(at
r
b
= 0.005) were subsequently incubated in 0.2 M NaCN, pH 8.3, to remove ~90% of the intrastrand cross-links, leaving ~85% ICLs (
47
; Table
1
). The DNA fragments containing ICLs were then separated from those containing no ICLs.
It has been suggested in this work (Fig.
4
) that extensive formation of long-range ICLs of cisplatin due to negative supercoiling is unlikely. In order to further support this suggestion, specifically for the (
Nde
I/
Hpa
I) fragment containing the ICLs formed by cisplatin in negatively supercoiled
DNA and used in the mapping experiments, this fragment was further cleaved by
other restriction enzymes, namely
Bgl
II,
Cla
I or
Hin
dIII. All these endonucleases cut only once within the (
Nde
I/
Hpa
I) fragment. Cleavage of the unmodified fragment by
Bgl
II,
Cla
I or
Hind
III yields 89 with 132, 102 with 119 or 158 with 63 bp fragments respectively.
The expected average size of loops closed by long-range ICLs formed in pSP73 DNA ([sigma] = -0.05) was estimated to be ~200 bp. Digestion of the (
Nde
I/
Hpa
I) fragment containing a long-range ICL between sites separated by ~200 bp by any of
Bgl
II,
Cla
I or
Hin
dIII should thus produce markedly shorter fragments, which would, however,
remain cross-linked. The electrophoretic mobility of these cross-linked fragments in 15% non-denaturing polyacrylamide gels should be approximately identical to that of the uncleaved (
Nde
I/
Hpa
I) fragment and it should be radically lower than that of the fragments obtained
by the same cleavage of the non-platinated (control) (
Nde
I/
Hpa
I) fragment. The (
Nde
I/
Hpa
I) fragments, control (non-platinated) sample and the sample used in the mapping experiments
containing the ICL formed in negatively supercoiled DNA (from which the
fragments containing no ICL were removed,
vide supra
) were cleaved under identical conditions by either
Bgl
II,
Cla
I or
Hin
dIII. The products of digestion of the interstrand cross-linked fragment were analyzed by electrophoresis in 15% non-denaturing polyacrylamide gels. They co-migrated with the fragments obtained by digestion of the
control, non-platinated (
Nde
I/
Hpa
I) fragment, i.e. they migrated markedly faster than the undigested (
Nde
I/
Hpa
I) fragments (shown for digestion by
Cla
I in Fig.
5
). Also, importantly, digestion of the interstrand cross-linked fragment was complete, leaving no material which would co-migrate with the undigested 221 bp fragment.
DISCUSSION
The results of this work demonstrate that DNA ICLs of cisplatin can be formed at
low levels of modification (
r
b
< 0.001) in cell-free medium with a considerably higher frequency in negatively
supercoiled DNA in comparison with relaxed or linear DNAs. The frequency of
ICLs in supercoiled DNA increases with decreasing level of platination. At low
levels of modification at 37oC, corresponding to
r
b
= ~10
-4
, after a modification reaction lasting 48 h the frequency of ICLs of cisplatin
in naturally supercoiled plasmid DNA was 30-40% of all adducts and approximately three times higher than in linear
DNA (Fig.
2
C). In addition, the kinetics of ICL formation in supercoiled DNA are markedly
faster than in linear DNA, which suggests that negative supercoiling makes
formation of ICLs of cisplatin easier.
Major structural reorganizations of B-DNA due to bifunctional lesions of cisplatin involve bending and unwinding
(
4
,
5
,
23
,
26
,
40
,
42
). There is no radical difference in bending induced by intrastrand cross-links and ICLs [32-35o for intrastrand adducts (
59
,
60
) and 20-45o for ICLs (
23
,
28
)]. In contrast, the unwinding induced by ICLs of cisplatin is ~80o (
23
,
28
), which is noticeably more than that induced by intrastrand cross-links (13 or 23o for 1,2 or 1,3-intrastrand adducts respectively;
40
). It has been shown that the free energy of the negatively supercoiled molecule
can drive reactions resulting in unwinding of the DNA double helix (
36
). Thus, a higher frequency of ICLs of cisplatin in negatively supercoiled DNA
is consistent with the fact that among reactions of cisplatin with DNA that
which results in ICLs requires the greatest extent of DNA unwinding.
A change in the topology of DNA could affect not only the frequency of formation of the individual types of cisplatin adducts, but also their preferences for particular base sequences in DNA. The analysis of
binding sites revealed (Fig.
6
) that ICLs were preferentially formed by cisplatin in both negatively
supercoiled and linear DNA at guanine residues and that these residues were
always contained in a GC base pair flanked by another GC pair. This result is
consistent with the view that ICLs formed by cisplatin in negatively
supercoiled DNA are preferentially formed between guanine residues in
neighboring base pairs. In addition, the results of the mapping studies (Fig.
6
) can also be interpreted to mean that in negatively supercoiled DNA ICLs are
formed by cisplatin between guanine residues in sequences containing both 5'-d(GC)-3' and 5'-d(CG)-3' sites. This is in contrast
to cisplatin modification of linear DNA, which results in ICLs between
neighboring guanine residues only in 5'-d(GC)-3' sites (
47
).
The results of replication mapping studies (Fig.
6
B) suggest that in negatively supercoiled DNA ICLs of cisplatin are not formed
at all 5'-d(GC)-3' and 5'-d(CG)-3' sites. This non-random distribution of
DNA binding sites at which cisplatin can form ICLs has not been investigated in
detail and will be addressed in future work. At present we only speculate that
this non-random distribution might be a consequence of the fact that the formation
of ICLs of cisplatin is affected by the base sequences flanking the 5'-d(GC)-3' and 5'-d(CG)-3' sites. In addition, ~15% of ICLs were
removed from the template DNA fragment during treatment with NaCN, which might
also contribute to the non-random distribution of termination sites due to ICLs. Thus, our results do
not exclude the possibility that a small fraction of ICLs of cisplatin can also
be formed at other sites in negatively supercoiled DNA than at those described
in this work.
DNA topogy is a significant factor in a number of genetic functions. Further
investigation of the effect of DNA topology on the formation of adducts of
cisplatin is warranted to define their role in the mechanism of toxicity of
this drug. In addition, the fact that the rate of formation of DNA ICLs of
cisplatin is dependent on the negative superhelical density could provide a
basis of a new assay for probing unrestrained tension in the winding of the DNA
double helix both
in vitro
and in living cells (
53
,
61
,
62
).
ACKNOWLEDGEMENTS
The research of Viktor Brabec was supported in part by an International Research
Scholar's award from the Howard Hughes Medical Institute. This work was also
supported by the Internal Grant Agency of the Academy of Sciences of the Czech
Republic (grant no. 504406) and the Grant Agency of the Czech Republic (grant
no. 307/96/0996). This research is also a part of the European Cooperation in
the field of Scientific and Technical Research (COST) Network (COST project
D1/0002/92). We thank to Professors Emil Palecek and Vladimír Kleinwächter for stimulating discussions and critical reading of the
manuscript.
4 Leng,M. and Brabec,V. (1994) In Hemminki,K., Dipple,A., Shuker,D.E.G., Kadlubar,F.F., Segerbäck,D. and Bartsch,H. (eds), DNA Adducts: Identification and Biological Significance, IARC Scientific Publication no. 125. International Agency for Research on Cancer, Lyon, France, pp. 339-348.
